Multiple instance convolutional neural network with modality-based attention and contextual multi-instance learning pooling layer for effective differentiation between borderline and malignant epithelial ovarian tumors

作者：

Highlights：

• For the first time, CNN was used for the automated BEOT/MEOT differentiation.

• The proposed MAC-Net outperforms several known MICNN methods.

• MA module can adaptively fuse multiple MRI modalities.

• C-MPL module can utilize contextual information between adjacent images.

摘要

•For the first time, CNN was used for the automated BEOT/MEOT differentiation.•The proposed MAC-Net outperforms several known MICNN methods.•MA module can adaptively fuse multiple MRI modalities.•C-MPL module can utilize contextual information between adjacent images.

论文关键词：Malignant epithelial ovarian tumors,Borderline epithelial ovarian tumors,Multiple instance learning,Multiple instance convolutional neural network

论文评审过程：Received 2 October 2020, Revised 1 September 2021, Accepted 7 October 2021, Available online 12 October 2021, Version of Record 15 October 2021.

论文官网地址：https://doi.org/10.1016/j.artmed.2021.102194